CN110308796A - A kind of finger movement recognition methods based on wrist PVDF sensor array - Google Patents
A kind of finger movement recognition methods based on wrist PVDF sensor array Download PDFInfo
- Publication number
- CN110308796A CN110308796A CN201910608346.2A CN201910608346A CN110308796A CN 110308796 A CN110308796 A CN 110308796A CN 201910608346 A CN201910608346 A CN 201910608346A CN 110308796 A CN110308796 A CN 110308796A
- Authority
- CN
- China
- Prior art keywords
- wrist
- data
- finger movement
- group
- sensor array
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/014—Hand-worn input/output arrangements, e.g. data gloves
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
The invention discloses a kind of finger movement recognition methods based on wrist PVDF sensor array, acquire the wrist movement signal of multichannel in real time using PVDF sensor array first, are pre-processed；Secondly the end-point detection based on short-time energy is carried out to wrist movement signal, obtains the cut-point of finger movement, and then extract and obtain the data of one section of finger movement；It is then based on short-time energy to encode the data of this section of finger movement, obtains the feature vector of single finger movement；Finally classified and identified using feature vector of the classifier to finger movement, obtains finger movement recognition result.Advantage of the invention is that, for user after wrist wears PVDF sensor array, hand is not necessarily to other auxiliary devices, can effectively identify elemental motion when mouse uses, such as index finger is clicked, middle finger is clicked, thumb is clicked, and provides a kind of new man-machine interaction mode for wearable mobile device.
Description
Technical field
The present invention relates to finger movement recognition methods field, specifically a kind of finger based on wrist PVDF sensor array
Motion recognition method.
Background technique
Important method one of of the gesture identification as human-computer interaction is widely used in Sign Language Recognition, prosthesis control, game behaviour
The fields such as work, remote control operation.
There are mainly two types of traditional Gesture Recognitions, and one is the gesture identification based on machine vision, it is current hand
Mainstream in gesture identification technology, but it is easy the interference by external environment, and cannot have shelter when in use.It is another
It is sensor-based gesture identification, it is mainly realized using inertial sensor with piezoresistance sensor, such as data glove,
But this method will limit the hand exercise of user；The gesture identification based on surface myoelectric sensor is also the heat of research simultaneously
Point, however it is high to the environmental requirement of skin surface when the wearing of this kind of sensor, and comfort level is low.
And sensor-based hand motion recognition method, often utilize multiple channels data, extract corresponding time domain,
The feature of frequency domain, time-frequency domain, using specific data classification algorithm, execution classification and identification.In practical application, wanting essence
It really identifies multiple and different gestures, then needs the sensor signal in multiple channels.And the increase of number of sensors, on the one hand cause
System complexity is promoted, and on the other hand also brings bigger noise jamming.Therefore how more gestures are identified in real time,
It is the major issue of current urgent need to resolve.
Summary of the invention
The object of the present invention is to provide a kind of finger movement recognition methods based on wrist PVDF sensor array, to solve
The sensor-based hand motion recognition method of the prior art is restricted the problem of larger, data processing is vulnerable to noise jamming.
In order to achieve the above object, the technical scheme adopted by the invention is as follows:
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: the following steps are included:
(1), using PVDF sensor array, the wrist movement signal of multichannel is acquired in real time；
(2), the short-time energy of the wrist movement signal and ambient noise of step (1) acquisition is calculated separately, and to step (1)
The wrist movement signal of acquisition is pre-processed, to remove ambient noise therein and filter out Hz noise；
(3), it is based on the short-time energy and pre-processed results that step (2) obtain, wrist movement signal is carried out based in short-term
The end-point detection of energy to obtain the cut-point of finger movement, and then is extracted and obtains finger movement data；
(4), the finger movement data obtained based on short-time energy to step (3) are encoded, and obtain single finger movement
Feature vector；
(5), the single finger movement feature vector that step (4) obtains is classified and is identified using classifier, obtained
Finger movement recognition result.
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: set PVDF
Sensor array is made of J PVDF sensor, and in step (2), defining x (n) is the more of J PVDF sensor n-th sampling
The wrist motion signal in channel, is represented by x (n)={ xj(n), j=1,2 ..., J }, x (n) is known as n-th group wrist motion letter
Number, while the short-time energy for setting n-th group wrist motion signal passes through formula as E (n)It can count
Calculation obtains E (n)；
If every time when power-up initializing, the ambient noise that is calculated of M group data that is acquired according to PVDF sensor array
Amplitude average value be xnois, wherein the total duration of M group data is no less than 3 seconds, while the short-time energy for setting background noise is average
Value is ε, passes through formulaε can be calculated.
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: step (2)
In, if the signal after n-th group wrist motion signal x (n) removal background is xr(n), then xr(n)=x (n)-xnoise,
The ambient noise in removal n-th group wrist motion signal x (n) is realized by the formula；
Defining H (z) is discrete transfer function used in power frequency filters, and passes through formula xf(n)=xr(n) * H (z) realizes filter
Except xr(n) Hz noise in completes pretreatment, and wherein * indicates that convolution algorithm, final pretreated signal are xf(n)。
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: it is described from
Dissipate transmission function
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: step (3)
Process it is as follows:
(3.1), Initialize installation acts starting mark device S=0, effective action marker F in endpoint detection subroutine
=0, short-time energy Low threshold E1=2 ε, short-time energy high threshold Eh=15 ε, by pretreated xf(n) end is input to E (n)
In point detection subprogram；
(3.2), judge whether E (n) is less than E1If being judged as YES, (3.5) are thened follow the steps；If being judged as NO, save
xf(n) and step (3.3) are executed；
(3.3), judge whether E (n) is greater than E1, and act whether starting mark device S is equal to 0, if being judged as YES, mark
Remember that this group of data are movement starting point A, and movement starting mark device S=1 is set, then executes step (3.6)；If being judged as NO,
Then follow the steps (3.4)；
(3.4), judge whether E (n) is greater than EhIf being judged as YES, effective action marker F=1 is set, executes step
(3.6)；If being judged as NO, effective action marker F=0 is set, is executed step (3.6)；
(3.5) judge whether effective action marker F is equal to whether 1 and E (n) is less than E1If being judged as YES, set
F=0, S=0 are set, and marks this group of data for movement terminating point B, and execute step (3.7)；If being judged as NO, step is executed
Suddenly (3.6)；
(3.6), one group of data X is removedf(n+1) with E (n+1), and step (3.1) are executed；
(3.7), the signal for the individual part note that will test is denoted as XAB, XAB={ xf(i) | i=nA..., nB, data
Length is nAB=nB-nA, to XABThe position for searching maximum value, is denoted as nmax, in the position Look-ahead p group data, search backward
Q group data are intercepted in (nmax- p) group~(nmax+ q) data between group, if there is no data position or data length when searching
Deficiency, as nmax- p < nAOr nmax+ q > nB, then in (nmax- p) group~nAGroup or nBGroup~(nmax+ q) group no data position mend
Zero, obtain the signal X for the individual part that action data length is lL={ xl(i) | i=p ..., q }.
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: step (3)
In, the high threshold E of the short-time energyh=(15 ± 2) × ε, the Low threshold E of short-time energyl=(2 ± 1) × ε, action data are long
L=p+q is spent,P is the length of rising edge, and q is the length of failing edge.
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: step (4)
In, to the signal X of the individual part after segmentationLIt carries out framing and calculates its short-time energy Ex, to short-time energy ExCarry out vectorization
Obtain the feature vector V of single finger movement；
To short-time energy ExData be rearranged for one-dimensional vector by the direction of row vector, the dimension of feature vector V is n
(V)=fn* J, wherein fnFor the frame length after framing, formula is as follows:
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: step (5)
In, classification is carried out to the feature vector V matrix of movement using the classifier of n (V) dimension input
A kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: described
Classifier is multilayer feedforward neural network.
Compared with prior art, advantage of the present invention are as follows:
Finger movement recognition methods proposed by the present invention based on wrist PVDF sensor array, carry out data acquisition with
When pretreatment, the influence of industrial frequency noise and ambient noise to signal amplitude, and the action data based on short-time energy are eliminated
The finger movement that can extract friction speed is extracted, in practical applications, user is after wrist wears PVDF sensor array, hand
Portion be not necessarily to other auxiliary devices, can to index finger click, middle finger click, thumb click etc. mouses use when elemental motion into
Row efficiently identifies, and provides a kind of new man-machine interaction mode for wearable mobile device.
Detailed description of the invention
Fig. 1 is system block diagram of the invention.
Fig. 2 is the method for the present invention flow diagram.
Fig. 3 is action signal end-point detection program flow diagram of the invention.
Fig. 4 is single channel action signal segmentation schematic diagram of the invention.
Fig. 5 is multichannel action signal encoding example figure of the present invention.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.
As shown in Figure 1 and Figure 2, user, which wears, carries out finger movement knowledge method for distinguishing based on PVDF sensor array, and step is such as
Under: the wrist movement signal of multichannel is acquired using multichannel PVDF sensor array (101) first and is pre-processed
(102)；End-point detection (103) are carried out to pretreated action signal secondly based on short-time energy, obtain single finger movement
Cut-point, segmentation obtain the data of single finger movement；Volume based on short-time energy is carried out to the data of single finger movement
Code (104), obtains the eigenmatrix of single finger movement, feature is classified and identified using classifier (105).
Wherein pretreatment (102) is one group to the sensor array acquisition multichannel constituted via J PVDF sensor
Wrist movement signal x (n) carries out DC component treatment, obtains xr(n)=x (n)-xnoise, further, to xr(n) it filters out
Hz noise obtains pretreated signal xf(n)=xr(n) * H (z), * indicate convolution algorithm；The wrist for acquiring multichannel is dynamic
When making signal, typical sample frequency is 1000Hz；And For used in power frequency filtering
Discrete transfer function；Define x (n)={ xj(n), j=1,2 ..., J } it is the J PVDF sensor passage wrist that n-th samples
Motor message, referred to as n-th group wrist motion signal, andFor the short-time energy of n-th group signal；Further
, the M group (M group total duration is no less than 3 seconds) of acquisition is utilized in each power-up initializing to wrist PVDF sensor array
The amplitude average value of background noise is calculated in dataAnd the short-time energy average value of background noise
Action signal end-point detection program flow diagram shown in Fig. 3, to pretreated signal xf(n) in accordance with the following steps
It is handled, obtained individual part XL, steps are as follows:
Step 200, Initialize installation act starting mark device S=0, effective action marker F=0, El=2 ε, Eh=15
ε；
Step 201, input xf(n), E (n)；
Step 202 judges E (n) < ElIf being judged as YES, step 209 is executed if being judged as NO and thens follow the steps 203；
Step 203 saves xf(n)；
Step 204 judges E (n) > ElAnd S=0 thens follow the steps 205 if being judged as YES, if being judged as NO, executes
Step 206；
Step 205 marks this group of data for movement starting point A, is n-thAGroup data, and S=1 is set, and executes step
211；
Step 206 judges E (n) > EhIf being judged as YES, 207 are thened follow the steps, if being judged as NO, is thened follow the steps
208；
Step 207, setting F=1, and execute step 211；
Step 208, setting F=0, and execute step 211；
Step 209 judges F=1 and E (n) < ElIf being judged as YES, 210 are thened follow the steps；If being judged as NO, execute
Step 211；
Step 210, setting F=0, S=0, and mark this group of data for movement terminating point B, it is n-thBGroup data, and hold
Row step 212；If being judged as NO, 211 are thened follow the steps；
Step 211 removes one group of data Xf(n+1) with E (n+1), and step (202) are executed
Step 212, the individual part that will test are denoted as XAB={ xf(i) | i=nA..., nB, data length nAB=
nB-nA, to XABThe position for searching maximum value, is denoted as nmax, Look-ahead pth group data, search q group data backward in the position,
Interception is in (nmax- p) group~(nmax+ q) data between group, if occurring no data position when searching or data length is insufficient, i.e.,
For nmax- p < nAOr nmax+ q > nB, then in (nmax- p) group~nAGroup or nBGroup~(nmax+ q) group the zero padding of no data position, obtain
The individual part X that regular length is 1L={ xl(i) | i=p ..., q }；
The wherein high threshold E of short-time energyh=(15 ± 2) × ε, the Low threshold E of short-time energyl=(2 ± 1) × ε, l=p+
Q,P is the length of rising edge, and q is the length of failing edge, particularly, is adopted when sample frequency is 1000Hz
The data length 1=194 of the movement collected.
It is shown in Fig. 4 be single channel action signal segmentation schematic diagram, show the original signal of continuous finger movement with
And the signal graph after signal segmentation.In Fig. 3, A is the starting point of a movement, and B is the terminating point of the movement.
The multichannel action signal encoding example figure shown in fig. 5 for being, to the signal X of the individual part after segmentationLIt carries out
Framing simultaneously calculates its short-time energy Ex, further, to ExIt carries out vectorization and obtains the feature vector V of individual part；In sampling frequency
When rate is 1000Hz, frame length w=5, frame moves inc=3, and the frame length after framing is fn=(l-w)/inc+1；Calculate framing in short-term
Energy are as follows:
Particularly, f when l=194n=64；
The dimension of feature vector V is n (V)=fn*J；Vectorization is to ExData be rearranged for by the direction of row vector
One-dimensional vector, formula are as follows:
Classifier using the input of n (V) dimension classifies to the eigenmatrix V of finger movement, wherein the classifier packet
It includes but is not limited to multilayer feedforward neural network.Using this method when being identified to finger movement, complete to finger movement
Identification in real time.
It is only in summary presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in essence of the invention
Within mind and principle, any modification, same replacement, improvement for being made etc. be should all be included in the protection scope of the present invention.
Claims (9)
1. a kind of finger movement recognition methods based on wrist PVDF sensor array, it is characterised in that: the following steps are included:
(1), using PVDF sensor array, the wrist movement signal of multichannel is acquired in real time；
(2), the short-time energy of the wrist movement signal and ambient noise of step (1) acquisition is calculated separately, and step (1) is acquired
Wrist movement signal pre-processed, ambient noise therein and filter out Hz noise to remove；
(3), it is based on the short-time energy and pre-processed results that step (2) obtain, wrist movement signal is carried out based on short-time energy
End-point detection, to obtain the cut-point of finger movement, and then extract and obtain finger movement data；
(4), the finger movement data obtained based on short-time energy to step (3) are encoded, and obtain the spy of single finger movement
Levy vector；
(5), the single finger movement feature vector that step (4) obtains is classified and is identified using classifier, obtain finger
Move recognition result.
2. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 1, feature
It is: sets PVDF sensor array and be made of J PVDF sensor, in step (2), defining x (n) is J PVDF sensor n-th
The wrist motion signal of the multichannel of secondary sampling, is represented by x (n)={ xj(n), j=1,2 ..., J }, x (n) is known as n-th group
Wrist motion signal, while the short-time energy for setting n-th group wrist motion signal passes through formula as E (n)
E (n) can be calculated in n > 2；
If every time when power-up initializing, according to the width for the ambient noise that the M group data that PVDF sensor array acquires are calculated
Value average value is xnoi, wherein the total duration of M group data is no less than 3 seconds, while setting the short-time energy average value of background noise as ε,
Pass through formulaε can be calculated.
3. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 2, feature
It is: in step (2), if the signal after n-th group wrist motion signal x (n) removal background is xr(n), then xr(n)
=x (n)-xnoise, the ambient noise in removal n-th group wrist motion signal x (n) is realized by the formula；
Defining H (z) is discrete transfer function used in power frequency filters, and passes through formula xf(n)=xr(n) * H (z) realization filters out xr
(n) Hz noise in completes pretreatment, and wherein * indicates that convolution algorithm, final pretreated signal are xf(n)。
4. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 3, feature
It is: the discrete transfer function
5. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 1, feature
Be: the process of step (3) is as follows:
(3.1), Initialize installation acts starting mark device S=0, effective action marker F=0 in endpoint detection subroutine,
Short-time energy Low threshold E1=2 ε, short-time energy high threshold Eh=15 ε, by xf(n) endpoint detection subroutine is input to E (n)
In；
(3.2), judge whether E (n) is less than E1If being judged as YES, (3.5) are thened follow the steps；If being judged as NO, x is savedf
(n), and step (3.3) are executed；
(3.3), judge whether E (n) is greater than E1, and act whether starting mark device S is equal to 0, if being judged as YES, label should
Group data are movement starting point A, and movement starting mark device S=1 is arranged, and then execute step (3.6)；If being judged as NO, hold
Row step (3.4)；
(3.4), judge whether E (n) is greater than EhIf being judged as YES, effective action marker F=1 is set, executes step
(3.6)；If being judged as NO, effective action marker F=0 is set, is executed step (3.6)；
(3.5) judge whether effective action marker F is equal to whether 1 and E (n) is less than E1If being judged as YES, F=is set
0, S=0, and mark this group of data for movement terminating point B, and execute step (3.7)；If being judged as NO, then follow the steps
(3.6)；
(3.6), one group of data X is removedf(n+1) with E (n+1), and step (3.1) are executed；
(3.7), the data for the individual part that will test are denoted as XAB, XAB={ xf(i) | i=nA..., nB, the data of movement are long
Degree is nAB=nB-nA, to XABThe position for searching maximum value, is denoted as nmax, in the position Look-ahead p group data, search q backward
Group data, intercept in (nmax- p) group~(nmax+ q) data between group, if there is no data position or data length when searching
Deficiency, as nmax- p < nAOr nmax+ q > nB, then in (nmax- p) group~nAGroup or nBGroup~(nmax+ q) group no data position mend
Zero, obtain the signal X for the individual part that action data length is lL={ xl(i) | i=p ..., q }.
6. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 1, feature
It is: in step (3), the high threshold E of the short-time energyh=(15 ± 2) × ε, the Low threshold E of short-time energyl=(2 ± 1) ×
ε, action data length l=p+q,P is the length of rising edge, and q is the length of failing edge.
7. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 1, feature
It is: in step (4), to the signal X of the individual part after segmentationLIt carries out framing and calculates its short-time energy Ex, to short-time energy
ExIt carries out vectorization and obtains the feature vector V of single finger movement；
To short-time energy ExData be rearranged for one-dimensional vector by the direction of row vector, the dimension of feature vector V be n (V)=
fn* J, wherein fnFor the frame length after framing, formula is as follows:
8. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 1 or claim 7, special
Sign is: in step (5), the classifier using the input of n (V) dimension classifies to the feature vector V matrix of movement.
9. a kind of finger movement recognition methods based on wrist PVDF sensor array according to claim 8, feature
Be: the classifier is multilayer feedforward neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910608346.2A CN110308796B (en) | 2019-07-08 | 2019-07-08 | Finger motion identification method based on wrist PVDF sensor array |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910608346.2A CN110308796B (en) | 2019-07-08 | 2019-07-08 | Finger motion identification method based on wrist PVDF sensor array |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110308796A true CN110308796A (en) | 2019-10-08 |
| CN110308796B CN110308796B (en) | 2022-12-02 |
Family
ID=68079291
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910608346.2A Active CN110308796B (en) | 2019-07-08 | 2019-07-08 | Finger motion identification method based on wrist PVDF sensor array |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110308796B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140314269A1 (en) * | 2011-09-14 | 2014-10-23 | Beijing Inforson Technologies Co., Ltd. | Device, system and method for recognizing action of detected subject |
| CN105022471A (en) * | 2014-04-23 | 2015-11-04 | 王建勤 | Device and method for carrying out gesture recognition based on pressure sensor array |
| CN105278699A (en) * | 2014-09-29 | 2016-01-27 | 北京至感传感器技术研究院有限公司 | Easy-wearable gesture identification device |
| US20170215768A1 (en) * | 2016-02-03 | 2017-08-03 | Flicktek Ltd. | Wearable controller for wrist |
| CN107045870A (en) * | 2017-05-23 | 2017-08-15 | 南京理工大学 | A kind of the Method of Speech Endpoint Detection of feature based value coding |
| CN107249080A (en) * | 2017-06-26 | 2017-10-13 | 维沃移动通信有限公司 | A kind of method, device and mobile terminal for adjusting audio |
| CN107392106A (en) * | 2017-06-26 | 2017-11-24 | 辽宁大学 | A kind of physical activity end-point detecting method based on double threshold |
| CN107766868A (en) * | 2016-08-15 | 2018-03-06 | 中国联合网络通信集团有限公司 | A kind of classifier training method and device |
| CN109144238A (en) * | 2018-05-14 | 2019-01-04 | 孙佳楠 | A kind of man-machine interactive system and its exchange method based on eye electricity coding |
| CN109567819A (en) * | 2018-11-28 | 2019-04-05 | 合肥工业大学 | A kind of wearable device based on PVDF sensor array |
| CN109976518A (en) * | 2019-03-12 | 2019-07-05 | 合肥工业大学 | A kind of man-machine interaction method based on PVDF sensor array |
-
2019
- 2019-07-08 CN CN201910608346.2A patent/CN110308796B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140314269A1 (en) * | 2011-09-14 | 2014-10-23 | Beijing Inforson Technologies Co., Ltd. | Device, system and method for recognizing action of detected subject |
| CN105022471A (en) * | 2014-04-23 | 2015-11-04 | 王建勤 | Device and method for carrying out gesture recognition based on pressure sensor array |
| CN105278699A (en) * | 2014-09-29 | 2016-01-27 | 北京至感传感器技术研究院有限公司 | Easy-wearable gesture identification device |
| US20170215768A1 (en) * | 2016-02-03 | 2017-08-03 | Flicktek Ltd. | Wearable controller for wrist |
| CN107766868A (en) * | 2016-08-15 | 2018-03-06 | 中国联合网络通信集团有限公司 | A kind of classifier training method and device |
| CN107045870A (en) * | 2017-05-23 | 2017-08-15 | 南京理工大学 | A kind of the Method of Speech Endpoint Detection of feature based value coding |
| CN107249080A (en) * | 2017-06-26 | 2017-10-13 | 维沃移动通信有限公司 | A kind of method, device and mobile terminal for adjusting audio |
| CN107392106A (en) * | 2017-06-26 | 2017-11-24 | 辽宁大学 | A kind of physical activity end-point detecting method based on double threshold |
| CN109144238A (en) * | 2018-05-14 | 2019-01-04 | 孙佳楠 | A kind of man-machine interactive system and its exchange method based on eye electricity coding |
| CN109567819A (en) * | 2018-11-28 | 2019-04-05 | 合肥工业大学 | A kind of wearable device based on PVDF sensor array |
| CN109976518A (en) * | 2019-03-12 | 2019-07-05 | 合肥工业大学 | A kind of man-machine interaction method based on PVDF sensor array |
Non-Patent Citations (3)
| Title |
|---|
| YAOHUI H等: "Piezoelectric poly(vinylidene fluoride) (PVDF) polymer-based sensor for wrist motion signal detection", 《APPLIED SCIENCES-BASEL》 * |
| 于海龙: "一种基于sEMG信号的手势识别方法研究", 《江西师范大学学报(自然科学版)》 * |
| 都悦来等: "基于传感器的手指运动姿态监测", 《黑龙江大学工程学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110308796B (en) | 2022-12-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105242779B (en) | A kind of method and mobile intelligent terminal of identification user action | |
| Savur et al. | Real-time american sign language recognition system using surface emg signal | |
| Yang et al. | Deep convolutional neural networks on multichannel time series for human activity recognition. | |
| CN107316067B (en) | A kind of aerial hand-written character recognition method based on inertial sensor | |
| Hisham et al. | Arabic Static and Dynamic Gestures Recognition Using Leap Motion. | |
| Lopes et al. | Hand/arm gesture segmentation by motion using IMU and EMG sensing | |
| Mohandes et al. | Prototype Arabic Sign language recognition using multi-sensor data fusion of two leap motion controllers | |
| WO1997020286A1 (en) | Method and system for handwriting recognition | |
| CN101354608A (en) | Method and system for implementing video input | |
| CN111089604B (en) | Body-building exercise identification method based on wearable sensor | |
| Ariesta et al. | A Survey of Hand Gesture Recognition Methods in Sign Language Recognition. | |
| He | Activity recognition from accelerometer signals based on wavelet-ar model | |
| Lamari et al. | Hand alphabet recognition using morphological PCA and neural networks | |
| CN106484082B (en) | Control method and device based on bioelectricity and controller | |
| CN111475030B (en) | Micro-gesture recognition method using near-infrared sensor | |
| CN112464738A (en) | Improved naive Bayes algorithm user behavior identification method based on mobile phone sensor | |
| Stiefmeier et al. | Fusion of string-matched templates for continuous activity recognition | |
| CN107346207B (en) | Dynamic gesture segmentation recognition method based on hidden Markov model | |
| CN109933202B (en) | Intelligent input method and system based on bone conduction | |
| Nandwana et al. | A survey paper on hand gesture recognition | |
| CN110308796A (en) | A kind of finger movement recognition methods based on wrist PVDF sensor array | |
| CN109634408A (en) | A kind of extended method of Hololens gesture identification | |
| CN112084898A (en) | Assembling operation action recognition method based on static and dynamic separation | |
| Kaur et al. | Deep learning for dynamic hand gesture recognition: Applications, challenges and future scope | |
| CN113057628A (en) | Inertial sensor based motion capture method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |